INDEX

Explanations

concepts related to truth, deception, and revealing secrets

oai_token-act-pair · gpt-4o-mini Triggered by @bot

like "reveal," "unearth," or "extract"

np_acts-logits-general · gemini-2.0-flash

reveal or uncover information

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

MigrationBuilder

-0.69

DeleteBehavior

-0.61

 autorytatywna

-0.57

RegistryLite

-0.56

padek

-0.51

asley

-0.51

χη

-0.49

ImageContext

-0.48

ViewFeatures

-0.46

<bos>

-0.46

POSITIVE LOGITS

 revealed

1.77

 reveal

1.73

 revelation

1.58

 reveals

1.52

 uncovered

1.52

 revealing

1.51

reveal

1.49

 révélé

1.48

revealed

1.48

 revelations

1.46

Activations Density 0.361%

concepts related to truth, deception, and revealing secrets

like "reveal," "unearth," or "extract"

reveal or uncover information

No Comments

No Known Activations